PCI-Express evolves into the news of the day from ATI and nVidia
We do not pretend that you learn to identify one or the other, which on the other hand is something that most already know how to do and that does not involve complications, but rather to introduce us to the world of the motherboard and its expansion slots.
The PCI is the most abundant in the plates, being able to find from only one, in mini-ATX up to six ATX. Its acronym stands for Peripheral Component Interconnect and works at about 33 Mhz base, these may vary if we make changes to the configuration of our motherboard.
Before continuing further, we will stop to rethink what 33 Mhz implies. Let’s imagine we have a 64-bit processor and our PCI at its corresponding speed.
1.33 Mhz is 33 million turns per second, that is, in a single second our PCI port is capable of 33 million steps.
2.64 bits are the data that the processor is capable of handling in each of these cycles.
3. If we multiply the 33 Mhz with the 64 bits, we obtain the number of bits, multiplied by one million, that our computer is capable of handling per second.
If we divide this previous figure by eight, remember that 8 bits are 1 byte, let’s find the number of bytes, again multiplied by one million, with which it works in one second.
5.As a kilobyte, kilo means a thousand, they are 1024 bytes, if we divide by 1024 it gives us the Kb.
6. Again we divide by 1024 to find the Mb, since obviously one megabyte is one thousand twenty-four kylobytes, or 1,048,576 bytes.
Finally, to find the figure in gigabytes per second, Gb / s, we divide by 1024 again.
Let’s see it applied to PCI:
33 x 64 = 2112 million bits per second
2112/8 = 264 million bytes per second
264 million / 1024 = 257,812 kilobytes per second
257,812 / 1024 = 252 megabytes per second
In this case we will not continue until we reach gigabytes since it is not necessary, although later we will.
As we can see, a PCI works with 252 Mb / s, a speed that can be enough for a modem, a sound card or even a network card, but if we want to work with, for example, a graphics card things get complicated.
It is not the same amount of data that a graphics card handles as a modem, so the AGP port was created.
This port that means Accelerated Graphics Port will sound familiar to all of us, we know where this name comes from since it is dedicated. But let’s get closer to him as it will be the basis for this news.
As we already know, the transfer capacity of the slot had to be increased, so instead of modifying the number of bits of the processor, which is extremely expensive, what was done was to work on the same bus that previously worked at 33 Mhz and now , in the AGP it does it to 66.
We will not repeat all the accounts, because as it is easy to imagine, 33 x 2 = 66, 252 Mb / sx 2 = 504 Mb / s. We could say that it is approximately half Gb / s.
It was in its day a very important step that made it possible to improve graphics cards in a significant way, but this was not enough, and more bandwidth was needed again. For this they again found the resource of multiplication.
Let’s say that nowadays the vast majority of graphics work with AGP 8X, although it has been preceded by 2X and 4X. Again we are going to skip the counts, and we will also go directly to 8X.
504 Mb / sx 8 = 4032 Mb / s not exactly 4 Gb / s, a little less, but we don’t care what we need it for.
Thus, a current graphics card is capable of working with about 4Gb for every second. It may seem to us that it is more than enough speed, or it may not seem just enough, but plenty, but nothing is further from the truth.
Graphics continue to require more and more bandwidth as their capabilities increase. And so we have come to need another change.
It is not worth looking for another multiplication, so a slot change was considered.
PCI-X is the name it received.
This also started with a speed of 66 Mhz, so if we have a 64-bit processor again, the speed will be exactly the same as AGP. The PCI-X 1.0 worked at 66 Mhz and 133 Mhz. Equivalent to AGP and AGP 2X.
The second revision of this one, the PCI-X 2.0, worked at 266 and 533 Mhz, that is to say, AGP 4X and AGP 8X.
Obviously we have not managed to improve the bandwidth at all, which is precisely what we were looking for. But there is always a previous step, an evolution we could say, until we achieve what we are looking for.
That is not to say that no benefits have been added to this PCI-X, such as ECC support, Error Correction Code. It is a code that is capable of detecting failures in the transmission of bytes, being able to correct them, if it is a single bit, or simply detect them if they are two.
But let’s not get into this technology that could take us pages and let’s continue the evolution until we get enough bandwidth.
This is the PCI-Express.
This one was able to carry out a multiplication that would correspond to an AGP 16X, that is, we would be talking about 8 Gb / s.
This is a very important figure because at the moment it offers enough width so that the new graphics feel comfortable. Although we do not forget to mention that not only the graphics are interested, because others, such as 10 Gb network cards or fiber optic cards, find the solution to their problems in this type of slot.
But since what comes to replace is the AGP, at least initially, we will focus on it and forget the other types.
ATI and nVidia are the producers of graphics chips that are currently in the high zone of the market. This enables them, and somehow forces them, to join this trend, both so as not to be left behind and to get their cards to perform at maximum power.
Therefore, from the beginning both houses expressed their interest in PCI-Express, nVidia in particular already did it in September of last year. He announced on his official website the reasons why he made this decision and began to work together with many other companies to reach the current situation.
But in the midst of all this, a lot happened, models were tested, samples were exhibited at computer shows, and perhaps most important of all, nVidia introduced the MXM. Mobile PCI-Express Module.
This is the portable version of the PCI-Express port as it is easy to imagine. The MXM turns out to be a revolution for laptops, as it will begin standardization as it happened in its day on desktops, it will facilitate the installation and uninstallation of graphics, which in current cases in which they are integrated into the motherboard is impossible, and, above all, it will improve laptop graphics cards with this new standard.
Today nVidia announces that it is working very closely with Intel on the very next entry of higher speed PCI-Express. For this, Intel has hundreds of GPUs (Graphics Processing Units) on its workbench that it will test and to which it will adapt as much as possible.
But we cannot forget ATI that today shows us its news.
On the one hand, the equivalent to MXM, which in this case is called AXIOM. On this one will mount the new graphics for ATI laptops, the X600, VPUs (Visual Processing Unit) prepared for the PCI-Express as it could not be otherwise and that will also be able to carry this technology from the hands of ATI to the laptops.
It is true that MXM and AXIOM collide with each other, but the same happened in many other cases and you always have to propose options to be able to choose the most convenient one. If one of them offers better options compared to the other or, simply, the graphics chips that are mounted on it are better accepted than those of the competitor, it will lead to the progressive choice of one or the other method.
On the other hand, ATI also presents today its graphics for PCI-Express-ready desktops. They are the X800, X600 and X300.
We know that the X600 is used on top of the AXIOMs, but they can also be used on a card for desktop computers.
The power of the VPUs increases as their number increases, the lower ones being the X300 and the upper ones the X800, although in the X300 it should be noted that they are the first graphics spark to use 0.11 micron technology. The other two top models require the 0.13 micron.
These graphics, in addition to working with these two technologies, use GDDR3 RAM, so that all the speeds it needs to improve performance are adapted to each other.
Manufacturers will begin to introduce their graphics cards based on these VPUs at the end of this month as expected by ATI.
In addition, and to finish now, the Canadian company increases the performance of these cards with the FireGL. The default settings, from highest to lowest throttle V7100, V5100, V3200 and V3100.
In this way, it also encourages the buyer with the new and revolutionary PCI-Express to do so with an improvement in performance on this, giving rise to excellent cards that will give their all.
We have finished, but we want to say that although the news is somewhat extensive, it seemed appropriate, taking into account the succession of events related to PCI-Express, that the reader is informed as much as possible of what this means and what are its consequences.